Peer Revenue Equity in science

Papers

Fresh from arXiv, typeset properly. Highlighted words trade on the market.

All of arXiv

Sunday 30 August

Artificial Intelligence

On the Instance Hardness as a Decision Criterion in TinyML Systems

TinyML includes the implementation of machine learning on devices with limited memory and computing resources. With the development of technology, AI systems continue to scale in terms of size and computational requirements. This forces researchers to adapt methods to be environmentally sustainable by designing techniques for reducing computational costs and energy consumption in inferring AI models, even in small devices. In this work, we present preliminary findings on a novel application of the tree depth prune instance hardness method to the TinyML system. The results indicate that threshold control can change energy consumption with limited classification quality changes. This method allows us to adjust classification accuracy, thereby influencing computational complexity and energy consumption for inference. We present a work in progress with initial results as a proof of concept.
Read the full abstract

Computers and Society

Verification-Time Dependency on a Disappearing Evaluator

AI governance and assurance often assume that a consequential model-mediated decision can be reconstructed or tested after the fact. That assumption may fail when the evaluator that produced the decision is no longer accessible in the same version and execution context. This paper develops three verification-time constructs derived from Execution Governance (EG) 3.0: Decision-State Commitment, Independent Verifiability, and Counterfactual Auditability. Independent reprocessing of released Study 2 artifacts reproduces two original within-family behavioural comparisons: 52.0% modal-decision reversal for Llama 3.1 8B versus Llama 3.3 70B (26/50) and 30.0% for GPT-OSS 20B versus GPT-OSS 120B (15/50). The corrected baseline establishes that these are within-family comparisons, not provider-established succession. Post-hoc re-pairing against Groq-designated migration paths yields 64.0% and 38.0% reversal, but these figures remain descriptive because the cross-family invocation parameters were asymmetric. A 22-event retirement census independently recomputes to median 16.45 months, mean 18.72 months, range 3.9-40.3 months, with 17/22 intervals below 24 months, while also showing that evaluator availability can differ by service surface. The joint contribution is an operational verification-time protocol and optional Verification-Time Preservation Package (VTPP) specifying what evidence to bind at authorization time, what a separately trusted verifier can substantiate later, how stability and paired counterfactual tests should be calibrated, and which semantic checks remain beyond JSON Schema validity. The protocol is downstream and non-authorizing: it does not alter the EG Core Formula, add a seventh live condition, or state jurisdiction-specific legal admissibility.
Read the full abstract

Quantum Physics

Exact Resource Laws for Passive Wavelength Routing in Entanglement Networks

Entanglement-based networks provide a scalable framework for multiuser quantum communication by passively routing spectrally correlated photon pairs across interconnected nodes. Several wavelength-allocation schemes have already been demonstrated experimentally, but these designs do not yet provide a general way to determine how spectral use, receiver load, repeated connections, fan-out, and optical loss constrain one another. Motivated by a field-deployed four-node network, we formulate wavelength assignment as an exact resource-optimization problem on the network connectivity graph. We consider uniquely paired wavelength channels and require that no user receive both channels of a pair. For one-sided fan-out, we obtain an exact optimization for arbitrary networks, and closed-form resource laws for complete networks and for networks with one excluded partner per user. Allowing both conjugate wavelengths to fan out changes the resource landscape. In complete networks, a balanced binary hierarchy uses the minimum number of spectral layers among nonredundant assignments while reducing the maximum receiver wavelength load to a quantity that grows logarithmically with the number of users. For eight users, the source-cost ordering depends on splitter transmission: changing the transmission can change which architecture requires less pair generation. We formulate a BBM92 rate model for evaluating the corresponding architectures. For a single continuous-wave broadband source, the occupied wavelength pairs are generated with fixed relative weights, while detector singles and accidental coincidences couple the link rates. The same resource laws can therefore be evaluated using the measured parameters of the hardware that implements the network.
Read the full abstract

All authors Ekta Panwar, Gilberto Borges, Saeid Salari, Kartik Kakade, Samgeeth Puliyil, Peter Rapčan, Mario Ziman, Djeylan Aktas

Computer Vision

Matrix-Game 3.5: Enhancing Real-Time Streaming Interactive World Models with Patch Memory

Interactive world models extend video generation from offline clip synthesis toward persistent simulation of interactive virtual worlds, enabling applications in games, robotics, embodied agents, and XR. Achieving stable long-horizon interactive generation, however, remains challenging, as the model must simultaneously preserve scene geometry, dynamic consistency, and camera control while supporting real-time autoregressive generation. Building upon Matrix-Game 3.0, we present Matrix-Game 3.5, as shown in Figure 1, which advances real-time interactive world generation toward geometry-aware and long-horizon consistent simulation through three key improvements. First, we propose a unified geometry-aware memory framework, whose patch-memory and tiled-PRoPE components introduce no additional learnable parameters, combining explicit 3D patch retrieval with projective camera conditioning to enable geometry-consistent camera control and faithful long-horizon scene recall. Second, we introduce a static-dynamic disentangled world representation that separately models static scene geometry and dynamic subjects, preserving both geometric consistency and subject identity throughout long-horizon generation. Third, we develop a two-stage progressive real-time distillation framework that converts a bidirectional diffusion model into a few-step causal generator through Perceptual Flow Matching and curriculum based Self-Rollout DMD, enabling minute-long real-time interactive generation. Extensive experiments demonstrate that, with a unified training corpus spanning Unreal simulation environments, open-world games, and internet videos, MatrixGame 3.5 achieves strong performance in long-horizon scene recall, precise camera control, subject consistency, prompt-driven world generation, and stable real-time open-world interaction.
Read the full abstract

All authors Runjia Qian, Zile Wang, Jihai Zhang, Kai Zou, Wei Yu, Jiaxing Li, Zexiang Liu, Yaokun Li, Fei Kang, Kaichen Huang, Mengyin An, Haobo Zhang, Biao Jiang, Jiahua Wang, Haofeng Sun, Yang Liu, Yangguang Li

Analysis of PDEs

Proof of the planar Khavinson-Shapiro conjecture

Let $\Omega\subset\mathbb{R}^2$ be a bounded domain whose boundary is a finite union of pairwise disjoint Jordan curves. We prove the planar Khavinson–Shapiro conjecture in this setting: if every polynomial on $\mathbb{R}^2$ agrees on $\partial\Omega$ with a harmonic polynomial, then $\partial\Omega$ is an ellipse and $\Omega$ is its bounded interior.
Read the full abstract

Systems and Control

Finite Sample Identification of Analytic Nonlinear Systems

This paper studies the identification of linearly parameterized nonlinear (LPN) systems. Although LPN systems share the same linear parameterization structure as linear systems, they are more challenging to identify. In particular, previous work has shown, through a counterexample based on a piecewise-affine system, that non-active exploration is generally insufficient for LPN system identification. In this paper, we consider LPN systems with real-analytic feature functions. We show that non-active exploration is sufficient for the identification of this class of systems by establishing non-asymptotic convergence rates of least-squares estimation and set-membership estimation. In addition, we provide counterexamples to show that non-active exploration may not be sufficient for system identification for non-real-analytic systems, even if those systems are infinitely differentiable. We present numerical experiments to further support and validate our theoretical results.
Read the full abstract

All authors Negin Musavi, Ziyao Guo, Geir E. Dullerud, Yingying Li

Machine Learning

Diffusion-Based Inverse Design of Dielectric Resonator Metasurfaces for Shaping Smart Electromagnetic Environments

Future wireless systems are expected to transform the surrounding space from a passive propagation medium into a smart electromagnetic environment, where engineered surfaces control wave propagation, support wireless sensing, and create programmable electromagnetic fingerprints. A key challenge in realizing this vision is the inverse design of metasurfaces for tailored electromagnetic propagation. While forward analysis evaluates the response of a known geometry, the inverse task starts from a prescribed scattering signature and seeks a physically realizable structure that produces it. This inverse task is inherently nonlinear and often high-dimensional, while candidate solutions may be non-unique and provide no direct indication of practical realizability. Here, we introduce a conditional diffusion framework for inverse design of dielectric resonator metasurfaces from target angular scattering patterns. Trained on T-matrix simulated geometry-response pairs, the model learns a conditional distribution of geometries instead of a deterministic mapping, enabling multiple candidate designs for the ill-posed inverse problem. The best generated metasurface achieves a mean percentage error of 1.39%, outperforming CMA-ES optimization (4.1% after 10 h) while requiring only about one minute for after-training inference. The model also produces lower error distributions than deterministic neural baselines for out-of-distribution spectra, highlighting the potential of diffusion models for efficient metasurface design.
Read the full abstract

All authors M. Tsukerman, K. Grotov, D. Vovchuk, P. Ginzburg

Information Theory

The generalized covering radii of Melas codes

The generalized covering radii have recently emerged as fundamental parameters of linear codes with applications to database linear querying. In this paper, we study the generalized covering radii $\rho_t(M(m,q))$ of Melas codes $M(m,q)$ over any finite field $\mathbb{F}_q$. We determine $\rho_2(M(m,q))$ for all $q$, and for a general $t \ge 3$, we prove that $\rho_t(M(m,q)) \in \left\{2t,2t+1\right\}$ for $q \in \{2,3\}$ and $\rho_t(M(m,q))=2t$ for $q \ge 4$ whenever $m$ is sufficiently large. These results extend recent work on the covering radius of Melas codes.
Read the full abstract

2608.29906 · 30 Aug · We have updated the references to reflect recent literature on this topic. In addition, we have clarified the relationship between our work and a related in-press article by another group, which addressed the binary Melas codes

Computer Vision

OrnaStyler: Ornament-Aware Latent Editing for Content-Preserving 3D Stylization

Text-guided style editing of 3D assets is essential for adapting existing objects to diverse visual aesthetics in digital content creation. Despite rapid progress in 3D shape modeling, faithfully stylizing an existing asset remains challenging when the desired stylization involves fine-grained structural ornamentation, which requires the model to preserve the source geometry and object identity, while coherently integrating new style-specific details. We propose OrnaStyler, a zero-shot framework for text-guided ornament-aware 3D stylization. Built upon rectified flow-based generative modeling, OrnaStyler introduces an inversion-guided editing strategy that recovers content-aware latent representations at both geometry and appearance levels in a staged manner to facilitate faithful editing. Our core idea is to explicitly model the spatial configuration of stylistic elements, thereby mitigating the fundamental tension between content preservation and style expression in the voxel space. Specifically, at the geometry level, we manipulate voxel representations through flow inversion to synthesize ornament-enhanced structures while preserving the spatial identity of the source asset. Then, at the appearance level, we introduce an adjacency-aware feature inpainting mechanism to harmonize newly generated ornaments with the original content, yielding coherent geometry-appearance integration. Our approach operates solely in the inference phase and enables selective editing over geometric augmentation or appearance stylization. Extensive experiments on both generated and real-world 3D assets against prior methods demonstrate that OrnaStyler achieves state-of-the-art editing performance in terms of content preservation, style fidelity, and overall visual realism. Code is available at: https://github.com/tomohiro0427/OrnaStyler
Read the full abstract

Computer Vision

Off-Manifold Refinement: Guiding Video Generators with a Frozen World Model

Modern video generators routinely fail at physical dynamics: objects float, trajectories violate gravity, contacts vanish. Standard denoising and flow-matching objectives fit visual data distributions but do not explicitly penalize such physical violations. Existing remedies can improve physical consistency, but typically add substantial inference or training cost. Candidate-selection methods generate and score multiple videos, while gradient-based world-model guidance repeatedly decodes and re-encodes intermediate estimates. Generator-internal refinement adds perturbation and re-denoising loops, whereas post-training requires curated data and additional optimization. We propose Off-Manifold Refinement (OMR), an inference-time method that instead injects world-model feedback directly into a single sampling trajectory. During scheduled middle ODE steps, we augment the generator velocity with the gradient of an adapter-space V-JEPA 2.1 surprise energy. This external correction can move the latent away from the uncorrected sampling trajectory and toward regions ranked as more physically plausible by the frozen predictor, after which the generator continues rendering from the corrected state. A small trained latent-to-embedding adapter keeps the gradient tractable at inference, and both the video generator and the world model remain frozen. On our fixed 400-prompt VideoPhy-2 detailed subset, OMR lifts the joint Semantic-Adherence-and-Physical-Commonsense metric from 47.0% to 52.0% (+5.0pp absolute, +10.6% relative) over the base Wan2.2-T2V-A14B sampler. On a separate fixed 50-prompt efficiency subset, it requires $1.71 \times$ the base runtime rather than the multiplicative cost of reward/search alternatives. Project page: https://itruonghai.github.io/omr.
Read the full abstract

Computation and Language

When Less is More: Understanding When Token Filtering Helps and Fails in AI-generated Text Detection

The rapid advancement of large language models (LLMs) has made AI-generated text detection increasingly critical. Existing zero-shot detectors assume that more token-level evidence leads to more reliable detection. However, our empirical study challenges this consensus: fewer tokens sometimes work better, retaining only 40% can yield optimal performance, yet this benefit is not universal. Using the Entropy Gap Score (EGS), we introduce top-$k$ cumulative probability filtering as a diagnostic probe. Across three representative settings, filtering exhibits strikingly different behaviors. We analyze EGS via typical set theory and quantify its dynamics through entropy calibration and distribution analysis. We find that filtering helps for weak source LMs, where low-entropy tokens are harmful, but fails for strong source LMs, where they are not notably harmful. Our work provides the first systematic analysis showing that some tokens are not merely uninformative but systematically harmful due to entropy miscalibration, revealing a two-sided trade-off in token-level detection.
Read the full abstract

Game Theory

Fair Division of Graphs: Beyond Traceability

In this paper, we study fair division problems in which resources are structured as graphs and agents must receive connected bundles. This connectivity requirement fundamentally alters the problem, making it significantly more challenging than its classical counterpart. We focus on the fairness notion of $\mathrm{EF1}_{\mathrm{outer}}$, where envy can be eliminated by removing at most one vertex whose deletion does not disconnect the bundle – a critical constraint for applications such as land division and network allocation. Our first result extends prior work by establishing the existence of $\mathrm{EF1}_{\mathrm{outer}}$ allocations for an infinite family of non-traceable graphs (that is, graphs that do not admit a Hamiltonian path), answering a central open question and generalizing Bilò et al.'s result for traceable graphs. We then make progress on a conjecture concerning the $\mathrm{EF1}_{\mathrm{outer}}$ spectrum of trees due to Chen and Zwicker. Finally, we complement our structural results with algorithmic insights, showing that deciding the existence of an $\mathrm{EF1}_{\mathrm{outer}}$ allocation is NP-complete even for binary additive valuations, thereby resolving an open complexity question. Taken together, our results deepen the connection between graph theory and fair division, and offer new tools for studying fairness in structured resource environments.
Read the full abstract

All authors Nicolas Bousquet, Frank Connor, Agnès Totschnig, Sébastien Zeitoun

Machine Learning

INTERVenE: Temporal-Abstraction-Interval Based Transformers for Short-Horizon Medical Event Prediction

Electronic Health Record (EHR) prediction models in the intensive care unit must learn from sparse and irregular measurements while preserving the clinical meaning of time and supporting transparent decision-making. We present INTERVenE, a family of Transformer architectures whose input is an interval-based, knowledge-based temporal abstraction (KBTA), a token stream of named clinical concepts (states, trends, events, contexts) drawn from a curated medical ontology, rather than an unnamed bin index or a raw measurement triplet. This naming layer is what we ask KBTA to do: it makes the model's per-token attributions resolve to clinical concepts by construction. INTERVenE offers two complementary variants: an auto-regressive decoder that generates future abstraction trajectories with a per-step risk readout (localizing when and after which events risk rises), and a bidirectional encoder for single-pass joint risk and time-to-event prediction. On 57,078 MIMIC-IV admissions, INTERVenE-Enc outperformed GRU-D, STraTS, and KarmaLego, reaching a support-weighted AUPRC of 0.672 ($+0.046$ over the strongest neural baseline, with non-overlapping 95% bootstrap CIs), AUROC of 0.901, and length-of-stay MAE of 44.4\,h. INTERVenE-Ar complements these predictions with generated abstraction trajectories and evolving risk scores, achieving AUROC of 0.854 and AUPRC of 0.588 under the harder generative readout. An alternative-discretization ablation preserved predictive performance, supporting KBTA's role as a clinically named attribution substrate rather than a prerequisite for accuracy.
Read the full abstract

Algebraic Geometry

Birational Automorphism Bounds for General-Type Foliations on Surfaces via Pluricanonical Indices

Let $\mathcal{F}$ be a canonical foliation of general type on a smooth projective surface $X$, and write $\mathrm{vol}(\mathcal{F}):=\mathrm{vol}(K_{\mathcal{F}})$. Let $G\subseteq\operatorname{Bir}(X,\mathcal{F})$ be a finite subgroup, and let $\mathcal{G}:=\mathcal{F}/G$ be the quotient foliation in the birational sense. For a canonical foliation $\mathcal{H}$, define its $r$-th pluricanonical index by \[ \delta_r(\mathcal{H}) := \min \bigl\{ m\in\mathbb{Z}_{>0} \mid h^0(mK_{\mathcal{H}})\geq r \bigr\}, \] where $\min\varnothing:=\infty$. For an arbitrary foliation, these indices are computed on any canonical birational model. If $\kappa(\mathcal{G})\geq0$, we prove \[ |G| \leq \begin{cases} 4\delta_1(\mathcal{G})\,\mathrm{vol}(\mathcal{F}), &\kappa(\mathcal{G})=0,\\[1mm] \displaystyle \frac{4}{3}\delta_2(\mathcal{G})\,\mathrm{vol}(\mathcal{F}), &\kappa(\mathcal{G})=1,\\[3mm] \delta_2(\mathcal{G})^2 \bigl(1+\delta_2(\mathcal{G})\bigr)\, \mathrm{vol}(\mathcal{F}), &\kappa(\mathcal{G})=2. \end{cases} \] Since $\operatorname{Bir}(X,\mathcal{F})$ is finite, one may in particular take $G=\operatorname{Bir}(X,\mathcal{F})$. When $\kappa(\mathcal{G})=0$ or $1$, the effective bounds $\delta_1(\mathcal{G})\leq12$ and $\delta_2(\mathcal{G})\leq42$ give \[ |G|\leq48\,\mathrm{vol}(\mathcal{F}) \qquad\text{and}\qquad |G|\leq56\,\mathrm{vol}(\mathcal{F}), \] respectively. The main new ingredient is a cluster formula for adjoint volumes, which yields index-dependent lower bounds for tangency-free foliated surface pairs whose underlying foliation has Kodaira dimension zero or one.
Read the full abstract

Computation and Language

REIGN: Refurbished Embeddings with Integrated Guidance Networks for Efficient Context-Length Scaling

Dense retrieval over long documents is expensive. Token-level encoders scale quadratically in sequence length, and most long-context embedding models reach 32K tokens only through architectural workarounds or by stretching billion-parameter LLMs. We propose REIGN (Refurbished Embeddings with Integrated Guidance Networks), a contrastively trained bi-encoder that operates on sequences of contextualised chunk embeddings from a frozen Guidance Network (GN) rather than on raw tokens. REIGN targets multi-chunk inputs, primarily for document-to-document retrieval; single-chunk inputs stay with the GN. Decoupling token-level processing from document-level reasoning, and caching the GN embeddings to disk, cuts per-document training cost by roughly four orders of magnitude relative to chunked Transformer fine-tuning. We also release a synthetic long-document retrieval benchmark for contrastive training and evaluation at long context lengths. Across an in-distribution Wikipedia benchmark, the LoCo out-of-distribution suite, and a real-world patent retrieval case study, REIGN matches dense long-context retrievers at smaller parameter budgets in each regime. A paired significance test puts it on par with models 1.6-4.3x larger on the patent task, and it stays within 0.65 nDCG@10 of a 20x-larger model on LoCo.
Read the full abstract

Materials Science

Cesium Clustering and Fluoroberyllate Network Disruption in FLiBe: A Total Scattering and Molecular Dynamics Study

Several next-generation fission reactor designs employ molten salts such as FLiBe (2LiF-BeF$_2$), with some concepts using fuel dissolved directly in the salt. During operation, fission products such as cesium will accumulate in the salt mixture, potentially leading to an evolution of the thermophysical properties underpinned by the atomic structure. To understand the structural perturbations in FLiBe with 5 mol% CsF, we conducted X-ray and neutron diffraction measurements, refined empirical potential structure refinement (EPSR) models against the experimental data, and compared the resulting structure with neural network molecular dynamics (NNMD) simulations. Comparisons of the EPSR and NNMD structures distinguishes features constrained by the scattering data from those that remain model dependent. The new Cs-bearing correlations account for the changes in the total structure factor and pair-distribution function, while the FLiBe correlations remain minimally altered. We find that Cs slightly disrupts the intermediate-range fluoroberyllate network, increasing the fraction of free fluorine ions, while the local coordination remains largely unchanged. The Cs ions within FLiBe cluster extensively, with BeF$_4^{2-}$ tetrahedra bridging neighboring cesium environments. In contrast to the minor structural perturbations in the liquid, the addition of 5 mol% CsF suppressed the formation of the crystalline Li2BeF4 phase at room temperature, with the phase appearing only above 180C upon heating. These experimentally constrained structural features provide a benchmark for atomistic models used to predict the behavior and properties of fission-product-containing FLiBe.
Read the full abstract

All authors Sean Fayfar, Rajni Chahal, Danny Wang, David J. Sprouster, Shravan Venugopal, Guiqiu Zheng, Dan Olds, Joerg Neuefeind, Stephen Lam, Boris Khaykovich

Computation and Language

When History Is Multimodal: Rethinking Context Management for Long-Horizon Agents

Long-horizon agents need a context manager to compress growing interaction histories into a bounded working context, via passive strategies or active strategies that decide how memory is accessed and reorganized. Meanwhile, prior optical-memory work mainly treats pixels as a dense codec for textualized histories, often presupposing that rendering context into optical memory incurs a significant performance drop relative to text, thus coupling this representation with SFT, self-distillation, or reinforcement learning to close this gap, leaving unresolved (i) how visual rendering performs as a context manager under a fair, controlled comparison, and (ii) whether this carrier offers a native advantage when history is inherently multimodal. In this paper, we formulate context management as a budget-constrained history transformation and introduce Visual Rendering (VR) as a representational context manager. Under a shared harness, policy model, trigger, and task domain, we evaluate VR on four text-centric and three multimodal benchmarks against four baselines (No Compression, Discard-All, Sliding Window, Summarization), finding visual memory is a natural carrier of native visual evidence. Building on this finding, we propose VERA (Visual Evidence-Retaining strategy for long-horizon Agents), a training-free context manager built on deterministic rendering with no exposed memory operations: on text-centric benchmarks it renders textual history as VR does, while on multimodal benchmarks it retains native visual observations instead of translating them into text. Across nearly all benchmarks, VERA cuts cumulative non-cache tokens by 31.5%-63.1% versus No Compression, matches existing managers on text-centric tasks, and achieves the highest accuracy among all baselines on multimodal tasks, supporting a modality-preserving view of long-horizon context management.
Read the full abstract

All authors Jiaqi Su, Cong Pang, Jiawei Hong, Tiankuo Yao, Zixuan Chen, Xin Lou, Lewei Lu

Robotics

EMERGE-Policy: A Robot Mind Emerges Beyond a Single Policy

A robot's effective “mind” need not reside in a single policy. It can emerge when specialized components perceive, reason, predict, act, verify, and remember within a shared orchestration process. EMERGE-Policy turns this perspective into a graph-structured agentic framework that coordinates both capability invocation and information exchange. A Main Agent retains task-level state within an active context window, while role-specific Sub Agents process perception, execution monitoring, verification, and memory consolidation in isolated contexts and return structured, task-relevant evidence. Role-specific contexts control information load by exposing only decision-relevant evidence to the Main Agent, while the functional Skill interface composes heterogeneous backends as Operational, Imagination, and Evaluation Skills. Criterion-grounded verification, textual failure diagnosis, and Branch Stack recovery provide localized correction, with token-aware external memory preserving task-relevant state. Together, their closed-loop interaction realizes the system-level policy captured by the name EMERGE-Policy. Without additional fine-tuning, we achieved outstanding performance on several public benchmark that have had a wide-reaching impact, and conducted a series of real robot experiments. These system-level results suggest that through the division of different functional sub-tasks among multiple agents and their concurrent collaboration, as well as the technical paradigm where the model is regarded as a skill and called within the framework, EMERGE-Policy can extend the robust robot policies beyond isolated runs.
Read the full abstract

All authors Zhirui Fang, Qingchi Yu, Ziyang Chen, Longfei Li, Haoran Ma, Keru Zhou, Xinrun Xu, Samith Va, Yuxuan Hu, Peixuan Song, Qiang Du, Bin Qian, Yongkang Deng, Xin Li, Yezhen Wang, Zhe Li, Hao Luo, Shuyan Li, Ziwei Wang, Weijian Deng, Xiu Li

Fluid Dynamics

Physics-Informed Kolmogorov-Arnold networks for viscoelastic fluid equations

Kolmogorov-Arnold Networks (KANs), inspired by the Kolmogorov Arnold representation theorem, provide an interpretable alternative to multilayer perceptrons (MLPs) by using learnable activation functions on edges rather than fixed node activations. We propose a Physics-Informed Kolmogorov-Arnold Network (PI-KAN) framework for solving forward problem of viscoelastic fluid equations, which arise in many complex fluid dynamics applications and are characterized by strong nonlinear coupling between fluid fields. For viscoelastic fluid equations, we adopt the generalized hydrodynamic model, which is well established in the field of dusty plasma. To evaluate the performance of the proposed framework for viscoelastic fluid, we consider benchmark problem based on the Taylor-Green (TG) flow and a modified Taylor-Green flow. We systematically investigate the effects of different network architectures, hyperparameters, and collocation point distributions on the accuracy and convergence behavior of PI-KANs for the range of viscoelastic parameter ($\tau_m = 1$–$20$). We also study the impact of random seed initialization on training outcomes. The obtained results provide useful guidance for the design and implementation of physics-informed Kolmogorov-Arnold networks (PI-KANs) in solving viscoelastic fluid equations
Read the full abstract

Mathematical Physics

Multiple Nonlinear Waves by (Quantum) Neural Networks: Checking the AI supremacy

To test the potential of so-called quantum physics-informed neural network (QPINN) technology for solving evolutionary partial differential equations, we consider the problem of multiple wave propagation in a weakly nonlinear medium described by the Korteweg-de Vries equation with periodic boundary conditions. Although this problem is solvable analytically, high-genus solutions could represent a rather challenging testbed for numerical methods. Therefore, our idea is to test whether a QPINN is sufficiently mature for application in this context by providing a set of indicators of merit and performing several numerical experiments.
Read the full abstract

Astrophysics of Galaxies

The Structural Abundance Crisis of Massive Galaxies in Current Cosmological Simulations

The internal structure of galaxies encodes the complex baryon cycle driven by gas accretion, star formation, and feedback, together with secular evolution and environmentally driven processes such as mergers and tidal interactions. We present a population-level census of massive nearby galaxies ($\log(M_\star/M_\odot) > 10$) by comparing Hyper Suprime-Cam Subaru Strategic Program observations with matched mock images from IllustrisTNG, EAGLE, and SIMBA. Using a consistent, like-for-like imaging pipeline, we construct structural abundance functions (SAFs) for key morphological parameters, revealing a structural abundance crisis. Across Sérsic index, concentration, size, and ellipticity, all simulations exhibit large, systematic discrepancies ($>5\sigma$; RMSE $\sim 0.2$–$1.8$\,dex), typically corresponding to abundance differences of factors of several. While individual simulations display diverse failures—underproducing or overproducing compact spheroids, extended or round galaxies —all underproduce highly flattened disks. Although TNG shows the closest agreement and SIMBA the largest offsets, this shared failure indicates that current models—despite matching global demographics such as the stellar mass function—do not uniquely constrain internal galaxy structure. Our results demonstrate that agreement in integrated observables can mask fundamental shortcomings in the modelling of mass and angular momentum redistribution. We therefore establish SAFs as a stringent, multidimensional, and observationally accessible benchmark for testing and calibrating next-generation galaxy formation models in the era of upcoming deep, wide-field surveys.
Read the full abstract

All authors Hassen M. Yesuf, Connor Bottrell, Luis C. Ho, Aaron S. G. Robotham, Sabine Bellstedt, Robin H. W. Cook, Lei Hao, Fengshan Liu

Machine Learning

Joint Spatiotemporal Spectral Neural Operators for Learning PDEs on Irregular Domains

Learning solution operators for partial differential equations (PDEs) on irregular and geometry-dependent domains remains a central challenge in scientific machine learning. While spectral methods provide strong inductive biases for modeling global interactions, they are typically limited to regular domains, and existing neural approaches often require domain warping, interpolation, or costly geometric embeddings. We introduce the Graph Spectral Neural Operator (GSNO), a neural operator that combines spatial graph spectral decompositions with temporal Fourier transforms through a unified space–time spectral kernel. This formulation enables globally coherent operator learning on non-Cartesian discretizations without domain warping or autoregressive rollouts. By replacing learned geometric embeddings with a graph Laplacian spectral basis, GSNO provides geometry-aware spectral learning with low parameter complexity. Across steady and unsteady PDE benchmarks on irregular and geometry-dependent domains, GSNO achieves strong accuracy with reduced runtime and parameter counts, while demonstrating robust zero-shot generalization across mesh resolutions and geometry families.
Read the full abstract

Computer Vision

MASQ: Mask-Aware Spatiotemporal Quantization for Unsupervised Skeleton Action Segmentation

Unsupervised skeleton-based temporal action segmentation is a crucial task for understanding human behavior in long untrimmed sequences. Recent approaches often rely on discrete quantization to discover action boundaries from motion representations. However, when spatial masking is introduced for representation learning, it can introduce representation ambiguity, while discrete quantization further amplifies small fluctuations in the latent space. The interaction between these two factors often leads to unstable code switching and severe temporal jitter near action boundaries.To address these limitations, we propose a novel Mask-aware Action Spatiotemporal Quantization (MASQ) framework. Our framework decouples the conflicting tasks of spatial feature inference and temporal smoothing.In the spatial dimension, we introduce a Joint-Level Structured Dropout (JLSD) mechanism that masks the entire temporal trajectory of selected joints, to encourage the model to learn discriminative inter-joint coordination patterns. In the temporal dimension, we design a mask-aware velocity loss that enforces motion consistency only on visible joints, that prevents gradient conflicts caused by masked signals and stabilizing temporal predictions. Extensive experiments on three widely used skeleton datasets, including HuGaDB, LARa, and BABEL, demonstrate that the proposed MASQ framework significantly outperforms existing state-of-the-art unsupervised methods. In particular, our model establishes a comprehensive and substantial leading advantage in the Mean over Frames accuracy.
Read the full abstract

All authors Xinyao Qin, Linxiang Peng, Youbao Ye, Di Yang, Jiangtao Wang

Computation and Language

En-ViMedNER: An English-Vietnamese Parallel Biomedical Corpus with UMLS Semantic Type Annotations

Biomedical Named Entity Recognition (NER) is fundamental to healthcare AI applications, including clinical decision support and medical information extraction. While corpora with Unified Medical Language System (UMLS) annotations, such as MedMentions, have driven progress in English biomedical NER, no comparable resource exists for Vietnamese. This paper presents En-ViMedNER, the first English-Vietnamese parallel biomedical NER corpus annotated with UMLS semantic types, which are language-neutral codes providing a shared cross-lingual label space and ensuring direct comparability with existing UMLS-based resources. The corpus contains 4,392 PubMed abstract pairs, 44,892 English-Vietnamese sentence pairs, and 202,949 aligned entity-mention pairs across 21 semantic types adapted from the MedMentions ST21pv dataset. To balance quality and scalability, we have constructed the corpus through automatic translation, expert post-editing, LLM-assisted label projection, and human verification and adjudication. We characterize En-ViMedNER as a large-scale silver-standard corpus with a human-audited and consensus-corrected mini-test subset. We evaluate En-ViMedNER in two settings: (i) Vietnamese-input/Vietnamese-output biomedical NER and (ii) English-input/Vietnamese-output cross-lingual NER. For Vietnamese NER, we benchmark Vietnamese-supervised encoder models, English-supervised multilingual encoder models, and prompt-based LLMs. The best model achieves an F1 score of 52.70 on the test set and 53.78 on the mini-test set. For cross-lingual NER, we benchmark encoder-decoder models and prompt-based LLMs. The best model achieves an F1 score of 45.44 on the mini-test set. We publicly release our corpus, corpus construction pipeline, and baseline models to facilitate future Vietnamese biomedical NLP research.
Read the full abstract

All authors Nhu Vo, Phuong Nguyen, Nu Uyen Phuong Le, Inigo Jauregi Unanue, Dung D. Le, Massimo Piccardi, Wray Buntine

Computation and Language

VibeJam: A User Study Platform for Web Development with Agents

Programming with AI is increasingly agentic, users prompt LLMs to directly edit their code and review the changes, with adoption growing especially for web development tasks. Despite this growth, most NLP work uses offline evaluation and lacks support for online studies, losing insights into how programmers truly use coding agents. We release VibeJam, a browser-based user study platform for users to collaborate with AI agents to develop websites. VibeJam enables agent customization and uses the open-source Aider agent by default, and to mirror downstream use, we add diff review, chat and plan modes, and live website previews. In a pilot study with 55 released, game-based website creation tasks, five experienced AI programmers rate our system as fun, simple, and resembling commercial tools, while 13 junior students use VibeJam to make websites of higher quality than agents in the same task. We open-source VibeJam to spur extensions and support studies on how coding agents can help users.
Read the full abstract

All authors Nishant Balepur, Connor Baumler, Valerie Chen, Eunsol Choi, Rachel Rudinger, Jordan Boyd-Graber

Machine Learning

Sensitivity-Constrained Neural Operators for Data-Efficient Forward and Inverse Modeling of Partial Differential Equation Systems

Neural operators provide fast surrogates for partial differential equation (PDE) solvers, but their reliability can degrade for high-dimensional spatial inputs and inverse or repeated inference. State-only training constrains solution values but not the learned input–output response. We study sensitivity-constrained neural operators (SC-NOs), which augment standard training with sampled solver-derived Jacobian supervision. Selected sensitivities from differentiable solvers or discrete adjoints are matched during training, allowing response information to be amortized across minibatches without imposing the full Jacobian at every update. We evaluate SC-NO on advection–diffusion and RANS–Spalart–Allmaras benchmarks, input-dimensionality scaling tests, long-horizon autoregressive rollout, and a shallow-water Tohoku tsunami source-inversion case. Sensitivity supervision improves forward prediction and yields larger gains in gradient-based inverse reconstruction of distributed fields. Scaling experiments show an improved accuracy–cost tradeoff for high-dimensional gridded inputs, while ablations indicate that state values and Jacobian information provide complementary supervision. In the tsunami case, SC-FNO reconstructs gridded seafloor deformation from sparse early gauge observations and forecasts subsequent wave propagation in a near-real-time proof-of-concept workflow. These results support sampled sensitivity supervision as a practical way to improve neural PDE surrogates when forward accuracy, inverse stability, robustness, and computational cost must be considered together.
Read the full abstract

All authors Abdolmehdi Behroozi, Chaopeng Shen, Daniel Kifer, Kathryn Lawson

Computation and Language

Check The Scoreboard: An Analysis of Scoring Schemes on Multiple-Choice Evaluation

Multiple-choice question answering (MCQA) benchmarks in NLP use number-right scoring (accuracy), but in educational testing, the scoring scheme, the combination of the response mode models follow and the rule for grading responses, is a key design choice that dictates which abilities to reward. We examine how alternatives to number right change what MCQA measures with six education-inspired schemes that assess abilities beyond accuracy: distractor elimination, abstention, confidence calibration, and self-correction. On LLM benchmarks, these schemes: 1) shift rankings of 31 LLMs beyond rephrased number right prompts; 2) better predict the LLMs users prefer in LLM Arena; and 3) reveal distinct model capabilities, like that GPT-5 rarely abstains and readily self-corrects, while weaker open-weight models often abstain and hesitate to eliminate choices. Given the benefits of alternative scoring schemes, we discuss ways to extend them to tasks beyond MCQA.
Read the full abstract

All authors Nishant Balepur, Paiheng Xu, Wei Ai, Eunsol Choi, Rachel Rudinger, Jordan Boyd-Graber

Machine Learning

Structural Hierarchy and Geometry in Molecular Representation Learning

Molecular self-supervised learning uses chemical structures to guide which molecular embeddings should be similar. We study whether explicitly encoding a molecule's Bemis-Murcko scaffold and using it to supervise the molecular embedding changes what the model learns. We further test whether this effect depends on the embedding geometry by comparing Euclidean and Lorentz contrastive objectives. Across two augmentation strengths, scaffold-supervised models consistently organize molecules according to both identical and structurally related scaffolds. The resulting embeddings also improve molecular property prediction on several tasks, while the exact gains depend on the predicted property. The effect of scaffold supervision on molecular organization is stronger under Lorentz objectives, but neither geometry provides a consistent overall advantage. These results show that explicitly teaching the relation between a molecule and its structural core can reliably shape the organization of molecular embedding space, while the extent of usefulness of this organization remains task dependent.
Read the full abstract

Symplectic Geometry

Existence of Reeb Chords from the Generation-Criteria

We give some existence results of Reeb chords using the $\textit{generation-criteria}$ of the Fukaya category. In particular, for $n\geq3$, we construct a Legendrian $\ell'\subset S^*\mathbb T^n$ such that every closed spin Legendrian $\ell\subset S^*\mathbb T^n$ either admits a self-Reeb chord or a Reeb chord joining it to $\ell'$.
Read the full abstract

Computation and Language

Improving Argument Saliency Coverage in Small LLMs for Long Legal Opinion Summarization via Sequence-Level Distillation

We show that sequence-level distillation from a capable long-context teacher model is a simple, annotation-free, and data-efficient strategy for improving argument saliency coverage in long legal opinion summarization, where small LLMs often struggle to retain the most salient argumentative content. Across student model sizes, distillation consistently surpasses tuning on expert-written summaries in our legal-opinion setting. We further demonstrate that most gains are achieved with as few as ~10 training summaries, highlighting the strong data efficiency of teacher-generated supervision. Finally, we find that summary distillation is sufficient for improvements: reasoning-chain distillation remains competitive with summary-only distillation, but provides marginal benefit when combined with summary supervision.
Read the full abstract